Are Rocket Internet copycats lean?
"My advantage is never that I'm the first. Never. My advantage is that we just build faster and better in more instances than anyone else."[1]
– Oliver Samwer, Founder of Rocket Internet
As I attended Oliver Samwer’s speech on the day of the Aardvark case, I couldn’t help reflect on whether I had applied the lean methodology during my summer in one of his companies. With numerous successful ventures, had Rocket applied the lean methodology or did focusing on execution and scaling with proven hypothesis (in a different country however) prevent Rocket Internet from following any sort of lean methodology?
Initial assumption - copycats don’t need lean principles?
My initial assumption was that by copying a successful idea in a different country, copycats do not need to prove key hypothesis such as customer demand or the feasibility of the business model. Therefore, an MVP was not needed and after a 3 month period where the Berlin IT team copied almost identically the website of their US counterpart the founders only had to focus on scaling.
Lean lessons from the clone factory
I soon realized that the US start-up that was copied was actually Rocket’s MVP. From this MVP, founders then needed to constantly iterate to achieve local product-market fit. While I was at Rocket, every change in the product followed the “build-measure-learn-scale” process that proved extremely efficient.
i. Build
The initial scope of an incremental change was usually extremely limited and would expand with each feedback loop. However, engineers usually spent no more than 24 hours building a new feature / test / pilot. Once it was launched, the team released up to one new version a day.
ii. Measurement
These new versions were monitored by a wealth of tools (i.e. scorecards, processes, etc…) that continuously test their incremental improvements. For example, each of the thousands of SEM keyword that was used to market the website was tracked with 3 to 5 KPIs and reported daily to the founder. The level of granularity at which each marketing dollar ROI was measured was quite impressive.
iii. Learn
Based on the results of the initial tests / pilots, management would try and identify the key learnings that would be incorporated in all further versions.
Identifying the limits – A locally optimized product that does not achieve a global optimum
Although Rocket ventures seemed to exhibit some characteristics of a lean venture, what struck me is the chaos between these build-measure-learn-scale loops. The overall feeling this experience left me with is that of multiple miniatures loops going on simultaneously and with limited coordination between each other (sometimes contradicting each other). Therefore, although each incremental change seemed to follow the lean principles the overall product development lacked both coordination and an established process / roadmap.
In addition, pressure being high from investors to crunch the timeline, it seemed as though “validated learnings” were only half-validated learnings. The data sets were usually minimal and therefore not always statistically significant enough to draw conclusions within a reasonable confidence interval.
Comments
Post a Comment